Effects of Independent Component Analysis on Magnetoencephalography Source Localization in Pre-surgical Frontal Lobe Epilepsy Patients

Abstract

Objective: Magnetoencephalography source imaging (MSI) of interictal epileptiform discharges (IED) is a useful presurgical tool in the evaluation of drug-resistant frontal lobe epilepsy (FLE) patients. Yet, failures in MSI can arise related to artifacts and to interference of background activity. Independent component analysis (ICA) is a popular denoising procedure but its clinical application remains challenging, as the selection of multiple independent components (IC) is controversial, operator dependent, and time consuming. We evaluated whether selecting only one IC of interest based on its similarity with the average IED field improves MSI in FLE. Methods: MSI was performed with the equivalent current dipole (ECD) technique and two distributed magnetic source imaging (dMSI) approaches: minimum norm estimate (MNE) and coherent Maximum Entropy on the Mean (cMEM). MSI accuracy was evaluated under three conditions: (1) ICA of continuous data (Cont_ICA), (2) ICA at the time of IED (IED_ICA), and (3) without ICA (No_ICA). Localization performance was quantitatively measured as actual distance of the source maximum in relation to the focus (Dmin), and spatial dispersion (SD) for dMSI. Results: After ICA, ECD Dmin did not change significantly (p > 0.200). For both dMSI techniques, ICA application worsened the source localization accuracy. We observed a worsening of both MNE Dmin (p < 0.05, consistently) and MNE SD (p < 0.001, consistently) for both ICA approaches. A similar behaviour was observed for cMEM, for which, however, Cont_ICA seemed less detrimental. Conclusion: We demonstrated that a simplified ICA approach selecting one IC of interest in combination with distributed magnetic source imaging can be detrimental. More complex approaches may provide better results but would be rather difficult to apply in real-world clinical setting. In a broader perspective, caution should be taken in applying ICA for source localization of interictal activity. To ensure optimal and useful results, effort should focus on acquiring good quality data, minimizing artifacts, and determining optimal candidacy for MEG, rather than counting on data cleaning techniques.

Keywords: MEG; frontal epilepsy; independent component analysis; interictal epileptiform discharges; magnetic source imaging; magnetoencephalography; source localization; spike.

Figure 1

Representation of the measures of accuracy. Dmin is the Euclidean distance expressed in mm between the source and the focus. Dmin dMSI is the distance between the maximum intensity (one single vertex) cortical map and the closest point of the focus. Dmin ECD is the distance between the ECD and the closest point of the focus. Proj_Dmin is the distance between the ECD projected to the closest cortical vertex and the closest point of the focus. Dmin_map is the distance between the border of the cortical map thresholded at 30% of the maximal intensity and the closest point of the focus.

Figure 2

Effect of ICA on ECD. (A) Dmin was not significantly different across ICA approaches as compared to no_ICA. Note that the y axis is in logarithmic scale. The median for all three methods was close to or lower than 1 cm. (B) Dmin of the dipole projection to the closest cortical point. No significant difference was found across ICA approaches as compared to no_ICA.

Figure 3

Effect of ICA on MNE. (A) Dmin was significantly lower (better) for No_ICA as compared to Cont_ICA and IED_ICA. (B) Dmin_map did not differ across approaches, probably due to a floor effect. (C) The spatial dispersion was significantly higher (worse) for both Cont_ICA and IED_ICA as compared with No_ICA. SD was also higher for IED_ICA as compared to Cont_ICA. *p < 0.05; **p < 0.001.

Figure 4

Effect of ICA on cMEM. (A) Dmin was significantly lower (better) for No_ICA and Cont_ICA as compared to IED_ICA. (B) Dmin_map was significantly worse for IED_ICA as compared to No_ICA. (C) The spatial dispersion was significantly higher (worse) for IED_ICA as compared with No_ICA and Cont_ICA. **p < 0.001.